Add plim and "Pr" to speech

[NSoiffer]

          I've never seen "plim" as a variant of "limit" before. I can add a rule for it. What would you like it to say? Maybe "p lim" for a terse reading and "probabilistic limit" otherwise?

Also, "Pr" is used for "probability" on that page. Do yo want that read as "Probability"? Reading "P" as probability is a bit dangerous without more context, but I think it is safe for "Pr".

Originally posted by @NSoiffer in #207 (comment)

[bhavyashah]

Another common probability notation is the double struck p.
"""
ℙ
"""
It should be read as "probability."

[bhavyashah]

Please let me know if this should be its own ticket. I think the underlying dilemma here is of note and relevant.
I was going to ask for double struck E to be read as "expectation." However, online sources say that it can be used for "Euclidean space, the expected value of a random variable, or a field in a tower of fields" (https://math.stackexchange.com/questions/304030/what-does-the-double-lined-capital-mathbbe-not-the-sigma-stand-for) and "Euclidean space" (https://mathworld.wolfram.com/Doublestruck.html). What is your vision for handling notations that are read differently in different contexts? My wishlist is: (a) add a user-friendly interface to make readings user-configurable, (b) put in some AI magic to gain contextual awareness, (c) implement subject areas (like statistics) where readings from that subject take precedence over alternative readings. For a more immediate solution, my thoughts are: either (a) use a neutral reading (like double struck E) unless a notation has one universal (or near-universal) reading, or (b) read as it would in the majority of cases (if it is established that double struck E is used for expectation more than half the time, then read it as expectation).

[NSoiffer]

Lots of things to answer:

Another common probability notation is the double struck p.

double struck P is often used for the set of all primes (similar to the set of all integers, rationals, etc).

I was going to ask for double struck E to be read as "expectation."

my thoughts are: either (a) use a neutral reading (like double struck E) unless a notation has one universal (or near-universal) reading,

In general I try to go with common readings where one is available, where "common" tends to be US high school and first year college as best as I know them. I try to use local context (global context isn't available to MathCAT, although in NVDA maybe the page can be scanned). When there isn't something obvious, I try to fallback on a neutral but simple reading. For example, for double struck E, it is spoken as "double struck E".

MathML 4 is going to introduce "author intent" so authors can influence the speech. It will be years before that is commonly used.

(c) implement subject areas (like statistics) where readings from that subject take precedence over alternative readings.

When author intent is not specified, my plan for better speech is for users to able to pick a subject area and get readings that are common for that subject area. That's already part of the dialog, but I haven't done any implementation so you can only select "General". "Geometry", "Probability & Statistics", and "Calculus" are three areas I will likely implement.

If is always possible to modify the rules, but that's not for the faint of heart. Modifying what is spoken for each character is pretty easy, but it will get overridden when you update. It might be possible to have a menu item that lets you define rules for characters. Defining rules for patterns is harder, although I did have a version of MathPlayer (never released) that had a GUI that allowed you to write a simple template that would override the normal rule.. Given that one often wants to restrict the rule from always firing by providing some tests, I'm not sure it really is usable. In the linked paper, you could remove special case rules but you'd have to write a new special case rule in text.

From your other comments, it appears you are taking classes in math. I would very appreciate if you could open an issue and suggest subject-specific readings where appropriate. Things like integrals don't need anything specific because there really is only one interpretation. However, "conditional probability" or "expected value" have ambiguous notations and knowing the subject area eliminates that. If you are willing to make suggestions based on past, current, or future classes, please open one issue per subject area. Thanks in advance.